Thallium Adsorption onto Illite - Environmental Science & Technology

Dec 29, 2017 - Eawag, Swiss Federal Institute of Aquatic Science and Technology, ... Institute of Biogeochemistry and Pollutant Dynamics, ETH Zürich,...
1 downloads 0 Views 2MB Size
Article pubs.acs.org/est

Cite This: Environ. Sci. Technol. 2018, 52, 571−580

Thallium Adsorption onto Illite Silvan Wick,†,‡ Bart Baeyens,§ Maria Marques Fernandes,§ and Andreas Voegelin*,† Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ü berlandstrasse 133, CH-8600 Dübendorf, Switzerland Institute of Biogeochemistry and Pollutant Dynamics, ETH Zürich, CH-8092 Zürich, Switzerland § Paul Scherrer Institute, CH-5232 Villigen PSI, Switzerland † ‡

S Supporting Information *

ABSTRACT: We investigated the adsorption of Tl+ onto purified Illite du Puy (IdP). Distribution coefficients (Kd) for trace Tl adsorption indicated a moderate pH-dependence from pH 2.5 to 11. Adsorption isotherms measured at Tl + concentrations from 10−9 to 10−2 M at near-neutral pH on illite saturated with Na+ (100 mM), K+ (1 and 10 mM), NH4+ (10 mM) or Ca2+ (5 mM) revealed a high adsorption affinity of Tl+ in Na+- and Ca2+-electrolytes and strong competition with K+ and NH4+. Cation exchange selectivity coefficients for Tl+ with respect to Na+, K+, NH4+, and Ca2+ were derived using a 3-site sorption model. They confirmed the strong adsorption of Tl+ at the frayed edges of illite, with Tl selectivity coefficients between those reported for Rb+ and Cs+. X-ray absorption spectra of Tl adsorbed onto Na-exchanged IdP indicated a shift from adsorption of (dehydrated) Tl+ at the frayed edges at low loadings to adsorption of (hydrated) Tl+ on planar sites at the highest loadings. Our results suggest that illite is an important adsorbent for Tl in soils and sediments, considering its often high abundance and its stability relative to other potential adsorbents and the selective nature of Tl+ uptake by illite.



TlIII in soil Mn-concretions and confirmed the importance of TlI sequestration by illite.12 Illite is a 2:1 type clay mineral with a basal spacing of 10 Å whose layers are held together by dehydrated K+ in the interlayers. Illite in soils may form by the weathering of muscovite or may be inherited from sedimentary parent material.25 Replacement of K+ at the edges of illite by larger and strongly hydrated cations like Ca2+ leads to an expansion of the interlayer spacing to 14 Å within the so-called frayed edge zone.26,27 The frayed edges of illite are known to have a very high adsorption selectivity for the weakly hydrated alkali metal cations K+, Rb+, and Cs+ and for NH4+. The adsorption of these cations onto illite has been extensively studied with respect to the availability of the nutrients K+ and NH4+28,29 and the retention of the radionuclide 137Cs in soils contaminated from nuclear accidents or in potential host rocks for radioactive waste storage.25,26,29−33 To describe the adsorption of Cs+ by illite in the presence of competing cations, a multi-site cation exchange model was developed by Brouwer et al. (1983)34 and generalized by Bradbury and Baeyens (2000)35 based on three types of cation exchange sites: (i) frayed edge sites (FES) with a very low adsorption capacity (0.25% of the cation exchange capacity (CEC) of illite) but very high affinity, (ii) type-II sites

INTRODUCTION Thallium (Tl) is a highly toxic trace element. It occurs in the environment predominantly as monovalent TlI and to a lesser extent as trivalent TlIII.1,2 Since the ionic radius of Tl+ (1.70 Å for 12-fold coordination) is similar to the ionic radius of K+ (1.64 Å),3 TlI can substitute K+ in K-bearing rock-forming minerals such as K-feldspars or micas,1,4,5 but is also found in rare minerals such as dorallcharite (Tl0.8K0.2Fe3(SO4)2(OH)6)6 and lanmuchangite (TlAl(SO4)2·12H2O).7 TlIII occurs in water as strongly hydrolyzing aqua cation Tl3+,1 forms poorly soluble avicennite (Tl2O3)8 and can be stabilized in Mn-oxides.9,10 Typical Tl contents in soils and sediments range between 0.1 and 1 mg/kg.11 High geogenic Tl contents of up to several thousand mg/kg may be found in soils formed on rocks hosting Tl-rich mineralizations.12,13 Anthropogenic Tl contamination can result from cement production,14 metal mining and smelting15,16 or coal burning.2,17 Because Tl+ is relatively soluble, it may be leached into the groundwater or taken up by plants and, via the food chain, may also pose a threat for human health.18,19 In soils and sediments, the solubility of Tl is controlled by the stability of Tl-bearing minerals and adsorption processes. Based on the geochemical association of Tl with micas and Mn-oxides,4,5,20 micaceous phyllosilicates such as illite and the phyllomanganate birnessite have long been assumed to be key adsorbents for Tl in soils and sediments.1,4,16,21−24 This assumption has been corroborated by a recent spectroscopic study on the speciation of Tl in a geogenically Tl-rich soil, which documented the occurrence of © 2017 American Chemical Society

Received: Revised: Accepted: Published: 571

August 31, 2017 November 25, 2017 December 13, 2017 December 29, 2017 DOI: 10.1021/acs.est.7b04485 Environ. Sci. Technol. 2018, 52, 571−580

Article

Environmental Science & Technology

solutions were prepared from TlNO3 in the required background electrolyte and were labeled with radioactive 204 Tl (β-emitter; T1/2 = 3.78 years; provided by RC TRITEC AG (Teufen, Switzerland); ICP-MS analysis of the tracer solution (0.6 MBq/g): 3.7% 204Tl, 94.3% 203Tl and 2% 205Tl). The background electrolyte was spiked with the desired total Tl concentration, and the pH of the solution was adjusted by either HNO3 or NaOH additions or by using organic pH buffers. Homoionic IdP suspensions were added to obtain the optimal final adsorbent concentrations, depending on the expected extent of Tl uptake. Subsequently, the tubes were closed and shaken end-overend at room temperature. Based on preliminary experiments on the kinetics of Tl+ uptake by IdP (details in Supporting Information; Figure S1), a reaction time of 14 days was used for the experiments described below. After reaction, phase separation was carried out by ultracentrifugation (1 h at 108,000 g (max), Beckman Coulter Avanti J30i High-Performance Centrifuge). Two 5 mL aliquots of the supernatant were radio-assayed for the remaining 204Tl activity using a liquid scintillation counter (Packard Tri-Carb 2750 TRILL) and the average of these two independent determinations of remaining activity was taken as one experimental data point. The pH was measured immediately after phase separation in the remaining supernatant. From the decrease in 204Tl activity and the known total Tl concentration (CTl,init, [mol/L]), the final dissolved Tl concentration in the supernatant (CTl, [mol/L]) was derived. From the decrease in dissolved Tl during reaction, the suspension volume (Vbatch, [L]) and the mass of illite (mIdP, [kg]), the amount of adsorbed Tl (QTl, [mol/kg]) was obtained. Distribution coefficients Kd (in [L/kg]) describing the ratio between dissolved Tl and adsorbed Tl were calculated according to eq 1:

(T2S) with an intermediate affinity and relatively high adsorption capacity (20% of CEC) and (iii) planar sites (PS) on the basal surfaces of illite with a low limited selectivity but high capacity (∼80% of CEC). Over time, cations adsorbed to FES can diffuse into the interlayer and replace K+, leading to their fixation in the illite structure, as observed for Cs+.26 Because of the similarities in dehydration energy and ionic radius between Tl+ and K+ and the occurrence of elevated concentrations of Tl in K-rich minerals including micas, Tl has been postulated to be effectively taken up by illite (or, more generally, micaceous clay minerals) in soils and sediments.1,4,12,16,21−24,36 However, the adsorption of Tl+ by illite has not been studied to date and cation exchange selectivity coefficients are lacking. In this study, we therefore investigated the cation exchange selectivity of Tl+ on purified homoionic Na-, K-, NH4-, and Ca-illite. We determined the pHdependence of Tl+ adsorption at trace concentration and the concentration-dependence at near-neutral pH. From the adsorption isotherms, cation exchange selectivity coefficients were derived based on the generalized 3-site Cs+ sorption model for illite. X-ray near edge structure (XANES) spectra at the Tl LIII-edge were collected on Na-illite with different Tl+ loadings to gain insight into the mechanism of Tl+ uptake.



MATERIALS AND METHODS Illite du Puy. The illite used for this work has been collected near Le Puy-en-Velay (France) and has previously been used in adsorption studies with different trace metals.37−39 Purified and conditioned Illite du Puy (IdP) suspensions used for the present work were prepared from the same illite source material and following the same procedure as described in detail in previous work.37 In a first step, IdP was conditioned to homoionic Na+ form (100 mM NaCl). Using the dialysis technique,37 initial Na-IdP suspensions were brought to Na+, K+, NH4+ and Ca2+ forms with different background electrolyte concentrations: Na-IdP (in 100 mM NaNO3), K-IdP (in 1 and 10 mM KNO3), Ca-IdP (in 5 mM Ca(NO3)2) and NH4-IdP (in 10 mM NH4NO3). The purified IdP has been reported to mineralogically consist of 88% illite and 12% sanidine.37 The small sanidine fraction was considered to be nonreactive with respect to Tl+ adsorption. A N2-BET surface area of 103 m2/g has been determined for purified IdP used in this study, in good agreement with a reported N2-BET surface area of 97 m2/g.37 The elemental composition of the purified IdP (48.4% SiO2, 20.4% Al2O3, 7.6% Fe2O3, 6.7% K2O, 3.9% MgO, 0.1% CaO and 0.9% Na2O) conforms with the compositional formula [Si6.98Al1.02O20(OH)4][Al2.44FeIII0.82Mg0.84][Ca0.02Na0.24K1.24] proposed by Poinssot et al. (1999)40 for another batch of Illite du Puy and with the description of Illite du Puy given by van Olphen and Fripiat (1979).41 Using the Cs isotope dilution technique42 with 204Tl instead of 137Cs, a Tl-CEC of 0.260 eq/ kg was determined for one of the purified IdP batches used in this study, in excellent agreement with the Cs-CEC of 0.263 eq/kg of the respective batch. General Setup of Adsorption Experiments. Batch adsorption experiments with homoionic IdP suspensions were conducted as a function of solution pH and Tl concentration in different background electrolyte (see SI Table S1 for an overview of the experimental conditions). All experiments were carried out in 40 mL polypropylene centrifuge tubes (Beckman Coulter). Stable 203/205Tl stock

Kd =

QTl C Tl

=

(C Tl,init − C Tl) C Tl

×

Vbatch mIdP

(1)

Each experiment was run in duplicate and the values were treated as independent data points. pH-Dependent Adsorption. pH adsorption edges were measured on Na-IdP (100 mM NaCl) and K-IdP (10 and 100 mM KNO3) in the pH range from ∼2.5 to ∼11.0. Total initial Tl concentration, that is, carrier 203/205Tl and tracer 204Tl, was 10−8 M. First, the Tl solution was added to the 40 mL centrifuge tubes. Subsequently, the pH was adjusted by either adding 0.1 M HNO3 (Merck Titrisol, for pH 4.0 to 5.0), 1 M HNO3 (Merck Titrisol, for pH 2.5−3.9), 0.1 M NaOH (Merck Titrisol, for pH 10.0−11.0) or 5 mL of preprepared organic buffer solutions (for pH 5.3 to 6.8: 0.015 M MES (2-(Nmorpholino)ethanesulfonic acid); for pH 6.9 to 7.8: 0.015 M MOPS (3-(N-morpholino)propanesulfonic acid); for pH 7.9 to 8.6: 0.015 M TRIS (Tris(hydroxymethyl)aminomethane); for pH 8.7 to 9.9: 0.015 M CHES (2-(cyclohexylamino)ethanesulfonic acid)). Finally, the IdP-suspension was added to achieve an IdP concentration of 2 g/L. Concentration-Dependent Adsorption. Adsorption isotherms were measured on Na-IdP (100 mM NaCl (low Tl concentrations) or NaNO3 (high Tl concentrations; to avoid precipitation of TlCl)), K-IdP (1 and 10 mM KNO3), Ca-IdP (5 mM Ca(NO3)2) and NH4-IdP (10 mM NH4NO3) for total Tl concentrations from 10−9 to 10−2 M. The solutions with different Tl concentrations were obtained by diluting a 10−2 M TlNO3 stock solution to different levels. The resulting 572

DOI: 10.1021/acs.est.7b04485 Environ. Sci. Technol. 2018, 52, 571−580

Article

Environmental Science & Technology

background NaCl concentrations of 100 mM (2.5 × 10−4 to 1.9 × 10−2 mol/kg), 95 mM (5.9 × 10−2 mol/kg) or 12 mM (2.4 × 10−1 mol/kg). After the 14-days reaction time, the Tl loaded illite was separated by ultracentrifugation and supernatant was decanted. The resulting wet paste was filled in acrylic glass sample holders covered with Kapton tape. Based on a water content of the illite pastes of ∼50% (by weight; after centrifugation), dissolved Tl was estimated to correspond to ∼3% of the total Tl in the most concentrated sample and to be insignificant in all other samples. A reference spectrum was collected on a 10 mM TlNO3 solution. The XANES spectra were recorded at the SuperXAS beamline (Swiss Light Source, PSI, Switzerland) at room temperature in fluorescence mode using a 5-element silicon drift detector. For data extraction and analysis, the software code Athena was used.46

solutions were labeled separately with 204Tl. The pH of each Tl solution was adjusted to ∼7.0 using 0.1 M NaOH or 0.1 M HNO3 (both Merck Titrisol). The solutions were pipetted in the 40 mL centrifuge tubes, followed by the addition of IdPsuspension to obtain an IdP concentration of 0.7 to 3.8 g/L. Final pH values after equilibration ranged between 6.1 and 6.9 (see Supporting Information (SI) Table S1). Adsorption Modeling. All adsorption isotherm data (5 isotherms; 230 data points) were used to derive cation exchange selectivity coefficients for Tl+-Na+, Tl+-K+, Tl+-NH4+, and Tl+-Ca2+ exchange based on the generalized 3-site cation exchange model for Cs+ adsorption onto illite of Bradbury and Baeyens (2000),35 that is, based on the respective site types and fractions. Cation exchange was treated according to the convention of Gaines and Thomas (1953)43 (i.e., assuming the activity of adsorbed cations to correspond to their equivalent charge fraction, eqs 2−4): n An + + BX m = AX n + Bm + (2) m KAB =



RESULTS AND DISCUSSION Trace Tl+ Adsorption from pH ∼2.5 to ∼11.0. Distribution coefficients (Kd) for trace Tl+ adsorption from pH ∼ 2.5 to ∼11.0 in 100 mM NaCl, 10 mM and 100 mM KNO3 background electrolyte are shown in Figure 1A. From 100 mM Na+ to 10 mM K+ electrolyte, a pronounced decrease in Kd values by about 1.5 log units can be observed over the entire pH range, indicating that K+ much more strongly interferes with trace Tl+ uptake than Na+. Kd values further decrease by about one order of magnitude when the K+ concentration increases from 10 to 100 mM. The constant decrease of the log Kd values over the probed pH range when the background electrolyte was varied from 100 mM Na+ over 10 mM K+ to 100 mM K+ clearly points to cation exchange as the dominant adsorption process over the probed pH range. The data further suggested that oxidation of Tl+ to poorly soluble or strongly binding Tl3+ did not detectably contribute to Tl removal from solution, since both Tl+ oxidation and Tl3+ precipitation or adsorption would strongly depend on solution pH rather than the type or concentration of the background electrolyte cation. From pH ∼2.5 to ∼11.0, Kd values in all three background electrolytes increased regularly by up to 1.3 log units. In the acidic region a smooth gradual decrease of the log Kd with decreasing pH cannot be explained by proton competition on exchange sites alone, because proton competition is only relevant at pH values ≤4 and would lead to a steep decrease in log Kd with further decreasing pH (as seen in model calculations for pH-dependent Cs+ uptake by illite).40 Specific adsorption of Tl+ at the variably protonated surface hydroxyl groups at the broken bond and edge sites (not FES) of illite platelets could contribute to the more pronounced increase of the adsorption affinity of Tl+ at alkaline pH.37 Based on the relationship between the hydrolysis constants of metal cations and their adsorption constants for coordination at illite edge sites reported in previous work37,47 and considering the pK1 of Tl+ of 13.2, adsorption of Tl+ onto these sites is expected to be weak at acidic to slightly alkaline pH values, but could become relevant at pH >10. Other factors that may contribute to increasing log Kd values with increasing pH could include changes in illite surface charge and structure. Finally, illite dissolution at acidic and alkaline pH could affect Tl adsorption, but was expected to be of minor relevance, considering that only 0.1% Illite du Puy has been reported to dissolve within 7 days at pH 3.040 and that dissolution rates at pH 4−10 are lower than at pH 3.48 Although the processes responsible for the observed pH-dependence of Tl+ adsorption could not be

(a An+)m × (NBXm)n (NAXn)m × (a Bm+)n

NAXn =

n × Q AX Q tot, X

n

NBXm =

(3)

m × Q BX Q tot, X

m

(4)

where A and B represent cations with charges n and m respectively, AXn and BXm the respective cations bound to the exchange site(s) X, KBA the cation exchange selectivity coefficient, aAn+ and aBm+ the activity of the free cations in solution (in [mol/L]), NA and NB the equivalent fractions of the exchangeable bound cations, QAXn and QBXm the amounts of the cations A or B bound to site X (in [mol/kg]) and Qtot,X the total concentration of site X (in [eq/kg]; X representing FES, T2S, and PS). The model was implemented in the geochemical equilibrium code PhreeqC using the thermodynamic database Minteq V4 and the Davies equation for the calculation of activity coefficients.44 For TlI, only the stability constants of the aqueous complexes TlCl, TlCl2−, and TlNO3 were included; oxidation of TlI to TlIII and precipitation of Tl-bearing phases were excluded. Similarly, only NH4+ and NH3 were included as dissolved N−III species, and oxidation of N−III, formation of any solution complexes or precipitation of solids was excluded. For parameter refinement, the PhreeqC model was set up in the software code PhreePlot,45 using the nonlinear least-squares (modified Levenberg−Marquardt) algorithm to obtain optimized parameters and their uncertainty. For parameter refinement, the sum of the squared residuals between experimental and calculated log Kd values was minimized, based on the total concentrations of Tl+, Na+, K+, NH4+, Ca2+, Cl−, and NO3− and the mass of homoionic illite in the suspensions. The fractions of FES (0.25% of CEC), T2S (20% of CEC), and PS (79.75% of CEC) on illite from the generalized 3-site Cs+ sorption model,35 and the experimental Tl-CEC (0.260 eq/kg) were used to calculate the individual site capacities, which were treated as nonadjustable parameters in the model (Qtot,FES = 0.00065 eq/kg; Qtot,T2S = 0.05200 eq/ kg; Qtot,PS = 0.20735 eq/kg). Tl LIII-Edge XANES Spectroscopy. For XANES analyses at the Tl LIII-edge, Tl was adsorbed to Na-IdP at Tl loadings from 2.5 × 10−4 to 2.4 × 10−1 mol/kg (50 to 49’000 mg/kg) at 573

DOI: 10.1021/acs.est.7b04485 Environ. Sci. Technol. 2018, 52, 571−580

Article

Environmental Science & Technology

Figure 1. (A) Log Kd of trace Tl+ adsorbed onto Na-illite (100 mM NaCl) and K-illite (10 and 100 mM KNO3) from pH ∼2.5 to ∼11.0. (B) Adsorption isotherms for Tl+ on illite at near neutral pH in 100 mM NaCl or NaNO3, 1 mM and 10 mM KNO3 electrolyte (diagonal gray dashed lines represent log Kd values of 4, 3, and 2 L/kg). (C) Adsorption isotherms for Tl+ on illite at near neutral pH in 5 mM Ca(NO3)2 and 10 mM NH4NO3. Data for 100 mM Na+ and 10 mM K+ from panel B are shown in gray for comparison.

Table 1. Cation Exchange Selectivity Coefficients for Tl+Na+, Tl+-K+, Tl+-NH4+, and Tl+-Ca2+ Exchange Refined from the Log Kd of Tl+ in Homoionic Na-illite, K-illite, NH4-illite, and Ca-illite Suspensions at near Neutral pH (Figure 2)a

identified in this study, it is important to note that the log Kd values in all three electrolytes vary by only ∼0.3 from pH 6 to 8, that is, that the pH-dependence of Tl+ adsorption onto illite at near neutral pH is weak. Tl+ Adsorption Isotherms in Different Background Electrolytes. In Figure 1B, the adsorption isotherms of Tl+ onto Na-IdP (100 mM NaCl or NaNO3) and K-IdP (1 and 10 mM KNO3) are shown. The comparison of the adsorption isotherms in 100 mM Na+ and in 10 mM K+ clearly demonstrate that K+ effectively reduces Tl+ uptake at low Tl concentrations (